Project progress forecasting method, progress forecasting device, progress forecasting program, and impact assessment model creation method
Patent Information
- Application Number
- JP2024162587
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-09-19
- Publication Date
- 2025-12-03
AI Technical Summary
Conventional project progress management systems struggle to accurately predict deviations between planned and actual progress, making it difficult to estimate future project timelines and take corrective measures.
A project progress prediction method and device that utilizes machine learning models to analyze past project data, calculate progress evaluation indices, and adjust predictions based on resource changes and inter-process impacts, enabling accurate forecasting of future project progress.
Enables precise prediction of project progress and deadlines at intermediate points, allowing for timely resource allocation and plan adjustments to meet project goals.
Smart Images
Figure 00000000_0000_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to a project progress prediction method, a progress prediction device, and a progress prediction program for predicting the future progress of an ongoing project. [Background technology]
[0002] Generally, when a project is being carried out, delays to the original plan (i.e., discrepancies between the progress plan and the actual progress) may occur. In such cases, the project manager needs to properly grasp the delay situation and take measures to resolve it.
[0003] Conventionally, in order to improve the accuracy of project progress management, a device is known that extracts information necessary for calculating delayed labor hours from a project progress information management file at any given time, stores the delayed labor hours calculated based on the work schedule and work results, and visualizes the history of the delayed labor hours (see Patent Document 1).
[0004] Also, in order to assist in the formulation of project schedules, a device is known which manages work process information that manages schedule information and actual result information for each work process in association with deviation information that indicates the reason for the difference between the schedule information and actual result information (the difference between the schedule and actual result for each work process), manages deliverable file information that has storage destination information where deliverable files generated in the work processes are stored in association with the work process information, and displays the work process information and deviation information, as well as the deliverable file information, on a display means (see Patent Document 2). [Prior art documents] [Patent documents]
[0005] [Patent Document 1] JP 2011-204098 A [Patent Document 2] JP 2002-32225 A Summary of the Invention [Problem to be solved by the invention]
[0006] In the conventional techniques described in the above Patent Documents 1 and 2, the progress of a project is grasped by man-hours (e.g., man-hours delayed from a plan). Therefore, in those conventional techniques, it is difficult to easily predict the progress of an ongoing project (e.g., how much deviation will occur between a specified date in the initial plan and the corresponding actual date), including future due dates (e.g., the midpoint or completion date of a process to be processed in the project).
[0007] Therefore, after careful consideration, the inventors of the present application have discovered that by taking into account the relationship between the planned period and the actual period at the midpoint of a construction project, etc. (the point at which future progress is predicted), future progress can be predicted through simple processing.
[0008] In view of the above background, an objective of the present invention is to provide a project progress prediction method, a progress prediction device, and a progress prediction program that are capable of predicting progress, including future deadlines, at the midpoint of a process in an ongoing project through simple processing. [Means for solving the problem]
[0009] In order to solve the above-mentioned problems, one embodiment of the present invention is a method for predicting progress of a project, the project including at least one process, comprising the steps of: acquiring progress plan data including a progress level of the process and a corresponding execution time; acquiring actual progress data including a progress level of a completed portion of the progress plan and a corresponding execution time; calculating an actual period from a reference point for the process to a specified time point at which a specified progress level is achieved based on the actual progress data; calculating a planned period from the reference point for the process to a time point at which a progress level identical to the specified progress level is achieved based on the progress plan data; calculating a progress evaluation index for the progress of the process based on the actual period and the planned period; and predicting the actual progress data after the specified time point based on the progress evaluation index.
[0010] According to this aspect, at a predetermined point (ie, a midpoint) of a process in an ongoing project, data on actual progress from the predetermined point (ie, progress including deadlines) can be predicted by simple processing.
[0011] In the above aspect, the progress evaluation index may be a ratio of the planned period to the actual period.
[0012] According to this aspect, the progress evaluation index can be calculated by simple processing based on the ratio of the planned period to the actual period.
[0013] In the above aspect, a fixed value may be used as the progress evaluation index.
[0014] According to this aspect, by using a fixed value as the progress evaluation index, the data of the progress results after a predetermined point in time can be predicted by simpler processing.
[0015] In the above aspect, the progress plan data is determined by a first machine learning model, and the first machine learning model may be a model that has learned the relationship between the progress of processes in past projects and actual progress including the corresponding execution timing.
[0016] According to this aspect, by using the first machine learning model, progress plan data can be acquired through simple processing.
[0017] In the above aspect, the prediction of the actual progress data includes a prediction of the completion date of the project, and if the completion date of the project exceeds a preset reference completion date, settings of resources required for executing the process after the specified time point are changed, and a revised evaluation index is calculated by correcting the progress evaluation index based on the changed resources, and the prediction of the actual progress data is made based on the revised evaluation index.
[0018] According to this aspect, even if the resources invested in each process at a specified point in time of a process in an ongoing project are changed, progress actual data from the specified point in time onwards, including the completion date, can be predicted through simple processing.
[0019] In the above aspect, the resources required to execute the process are determined by a second machine learning model, and the second machine learning model may be a model that has learned the relationship between the resources invested in the process in past projects and the construction period of the process after the resources are invested.
[0020] According to this aspect, by using the second machine learning model, the resources required for each process (that is, the changed resources) can be acquired through simple processing.
[0021] In the above aspect, the project includes, as the at least one process, a preceding first process and a subsequent second process, and acquires information on at least one first milestone in the first process and information on second milestones in the second process that correspond in time to each of the first milestones, acquires an impact assessment model that represents the influence on the progress assessment index of the second process of a relationship between data on each of the first milestones in the progress actual data of the first process and data on the corresponding second milestone in the progress actual data of the second process, calculates a revised evaluation index by correcting the progress evaluation index of the second process based on the impact assessment model at the specified time point, and predicts the progress actual data of the second process from the specified time point onwards based on the revised evaluation index.
[0022] According to this aspect, when a project includes multiple processes (first process, second process), data from a specified point in time (midpoint) of the second process onward is predicted based on a modified evaluation index modified on the basis of an impact assessment model that represents the impact between the processes. Therefore, in the subsequent second process, which is affected by the progress performance of the preceding first process, data on progress performance from a specified point in time onward can be predicted by simple processing.
[0023] In the above aspect, it is preferable that data on the progress of each of a plurality of completed projects is accumulated, and the first milestone and the second milestone are determined based on the accumulated data on the progress of each of the completed projects.
[0024] According to this embodiment, the first and second milestones can be determined by simple processing by using data on the progress results of processes in a plurality of completed projects.
[0025] In the above aspect, each of the first milestones and each of the second milestones may be determined based on the progress data of a project among the plurality of completed projects which has the shortest construction period for the second step.
[0026] According to this embodiment, appropriate first and second milestones can be determined by a simple process.
[0027] In the above aspect, it is preferable that data on the progress results of each of a plurality of completed projects is accumulated, and the impact assessment model is determined based on the accumulated data on the progress results.
[0028] According to this aspect, a model that represents the influence of the progress of a preceding first process on the progress evaluation index of a subsequent second process can be easily obtained based on data on the actual progress of processes in multiple completed projects.
[0029] In the above aspect, the impact assessment model may be a third machine learning model that learns a relationship between the accumulated progress achievement data and the progress assessment index of the second process.
[0030] According to this aspect, by using the third machine learning model, the corrected evaluation index in the second step can be calculated by a simple process.
[0031] In the above aspect, the revised evaluation index may be calculated based on a first milestone period from a reference point to each of the first milestones, a second milestone period from the reference point to each of the second milestones, and a difference between each of the first milestone periods and each of the corresponding second milestone periods.
[0032] According to this embodiment, the corrected evaluation index can be calculated by a simple process based on each of the first milestone periods and each of the second milestone periods and the difference between them.
[0033] In the above aspect, if a delay occurs in the timing of the first milestone in the actual progress data of the first process, the timing of the corresponding second milestone in the progress plan data of the second process may be changed to a time after the timing of the first milestone that was delayed.
[0034] According to this aspect, if a delay occurs in the first milestone in the first process, it is possible to avoid the execution timing of the second milestone in the second process being unnecessarily earlier than the first milestone.
[0035] In order to solve the above-mentioned problems, one aspect of the present invention is a progress prediction device having a processor that executes processing for predicting the progress of a project, the project including at least one process, the processor acquires progress plan data including the progress of the process and its corresponding execution timing, acquires actual progress data including the progress of completed portions of the progress plan and their corresponding execution timing, calculates an actual period from a reference point for the process to a specified point in time when a specified progress level is achieved based on the actual progress data, calculates a planned period from the reference point for the process to a point in time when a progress level identical to the specified progress level is achieved based on the progress plan data, calculates a progress evaluation index for the progress of the process based on the actual period and the planned period, and predicts the actual progress data after the specified point in time based on the progress evaluation index.
[0036] According to this aspect, at a predetermined point (ie, a midpoint) of a process in an ongoing project, data on actual progress from the predetermined point (ie, progress including deadlines) can be predicted by simple processing.
[0037] In order to solve the above-mentioned problems, one embodiment of the present invention is a progress prediction program that causes a computer to execute a process for predicting the progress of a project, the project including at least one process, and causes the computer to obtain progress plan data including the progress of the process and its corresponding execution timing, obtain actual progress data including the progress of a completed portion of the progress plan and its corresponding execution timing, calculate an actual period from a reference point for the process to a specified point in time when a specified progress level is achieved based on the actual progress data, calculate a planned period from the reference point for the process to a point in time when a progress level identical to the specified progress level is achieved based on the progress plan data, calculate a progress evaluation index for the progress of the process based on the actual period and the planned period, and predict the actual progress data after the specified point in time based on the progress evaluation index.
[0038] According to this aspect, at a predetermined point (ie, a midpoint) of a process in an ongoing project, data on actual progress from the predetermined point (ie, progress including deadlines) can be predicted by simple processing. Effect of the Invention
[0039] According to the above aspect, at the midpoint of a process in an ongoing project, it is possible to predict progress including future deadlines through simple processing. [Brief description of the drawings]
[0040] [Figure 1] Overall configuration diagram of a project progress forecasting system according to a first embodiment [Diagram 2] Functional block diagram of a progress prediction server according to the first embodiment [Diagram 3] An explanatory diagram showing an example of a progress plan for one process in a project [Figure 4] Diagram of SPI value calculation [Diagram 5] Diagram of progress data forecast [Figure 6] FIG. 13 is an explanatory diagram of data on progress results after resource change in a process according to the first embodiment; [Figure 7] FIG. 1 is a flow diagram showing a flow of progress prediction processing by a progress prediction server according to a first embodiment; [Figure 8] Functional block diagram of a progress prediction server according to a second embodiment [Figure 9] An explanatory diagram showing the process of matching milestones in the progress plans of two processes included in a project [Figure 10] Diagram showing the impact of a previous project on a subsequent project [Figure 11] FIG. 13 is an explanatory diagram of data on progress results after resource change in a process according to the second embodiment; [Figure 12] FIG. 11 is a flow diagram showing a flow of progress prediction processing by a progress prediction server according to a second embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0041] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a project progress forecasting method, a progress forecasting device, and a progress forecasting program according to an embodiment of the present invention will be described with reference to the drawings.
[0042] (First embodiment) FIG. 1 is an overall configuration diagram of a project progress prediction system 1 according to a first embodiment. FIG. 2 is a functional block diagram of the progress prediction server 2 shown in FIG. 1. FIG. 3 is an explanatory diagram showing an example of a progress plan for one process in a project. FIG. 4 is an explanatory diagram regarding calculation of an SPI value. FIG. 5 is an explanatory diagram regarding prediction of progress performance data. FIG. 6 is an explanatory diagram regarding progress performance data after a resource change.
[0043] Note that the data values (i.e., the shapes of the lines or curves represented by a set of data) shown in the graphs of Figures 3 to 6 are shown for the purpose of making the present embodiment easier to understand, and do not necessarily match the data values (shapes of the lines or curves) in actual projects.
[0044] 1, the progress prediction system 1 includes a progress prediction server 2 (an example of a progress prediction device) that performs processing to predict future progress of an ongoing project, and a plurality of user terminals 3 that are used by users such as project managers. Each user terminal 3 is capable of communicating with the progress prediction server 2 via a network 4 such as the Internet. Each user terminal 3 is composed of a computer such as a PC, a tablet terminal, or a smartphone.
[0045] In this embodiment, an example in which the progress prediction system 1 is applied to a construction project will be described. However, the progress prediction system 1 (or the progress prediction server 2) may be applied to any other project. Also, this embodiment will be described focusing on one process included in the project.
[0046] The progress prediction server 2 includes a control unit 11, a storage unit 12, and a communication unit 13, as shown in FIG.
[0047] In the control unit 11, a progress plan data acquisition unit 21 acquires data on the progress plans of the processes included in the project.
[0048] For example, as shown in FIG. 3, the progress plan data includes data on the progress level of the process to be processed (see the vertical axis) and the corresponding execution time (see the horizontal axis). In this embodiment, the execution time is the number of days elapsed from a reference point (e.g., the project start date) when the corresponding progress level is achieved. In FIG. 3, the time from the project start date (period=0 days, progress level=0%) to the completion date (period=X E The data shows the progress plan up to the date of completion (progress rate = 100%). In a construction project, the reference point may be, for example, the date of design completion, the date of completion of confirmation of the design drawings by the client, or the date on which materials required for construction arrive at the construction site.
[0049] In a construction project, the progress of a process can be expressed, for example, by the weight, length, or volume of a completed structure (including a partially completed structure). For example, in a foundation construction process, the progress can be determined by the volume of concrete actually used (the ratio of the amount used) to the total volume of concrete planned to be used. Note that for projects other than construction projects, the progress of a process can be determined similarly using a quantified progress indicator.
[0050] The progress plan data acquisition unit 21 can acquire data of the progress plan based on performance data of a plurality of projects carried out in the past (hereinafter referred to as "past performance data"). As these past projects, it is preferable to select projects that are the same as or similar to the project to be processed and that have been normally executed without delays exceeding an allowable range. In particular, it is preferable for these past projects to be performance data of projects with the shortest time from start to completion (construction period) without the addition of resources. The data of the progress plan may be the same as the performance data of the past projects, or may be the performance data in which a part of the performance data has been modified by the user. The past performance data is stored (or accumulated) in advance in the storage unit 12 as part of the project data 28.
[0051] Furthermore, the progress plan data acquisition unit 21 may acquire progress plan data using a progress plan acquisition model 31 (an example of a first machine learning model) which is a learning model obtained by machine learning. The progress plan acquisition model 31 is obtained by performing machine learning using teacher data indicating the relationship between the progress of each process in the past performance data and the corresponding execution time. The progress plan acquisition model 31 is stored in advance in the storage unit 12 as one of the learning models 30 that the progress prediction server 2 can use.
[0052] If there is no past performance data for a project that is the same as or similar to the project to be processed, the progress plan data may be newly created for the project to be processed. In this case, the progress plan data may be created based on the predicted progress and the corresponding execution time for the processes included in the project to be processed.
[0053] The progress plan data acquisition unit 21 can transmit screen data relating to the progress plan as shown in Fig. 3 to the user terminal 3 in response to a request from the user terminal 3. This allows the user to check the progress plan of the project from the screen relating to the progress plan displayed on the display of the user terminal 3. In addition, screen data relating to a plurality of progress plan candidates may be transmitted from the progress prediction server 2 to the user terminal 3, and the user may determine progress plan data from among the progress plan candidates by operating the user terminal 3.
[0054] The progress result data acquisition unit 22 sequentially acquires progress result data of the completed portion of the progress plan of the process to be processed for the ongoing project. The progress result data includes data on the progress level of the process to be processed and the results of the corresponding execution time, similar to the progress plan data. The acquired progress result data is sequentially stored in the storage unit 12 as part of the project data 28. Note that at least a part of the progress result data may be data input by the user.
[0055] An actual SPI (Schedule Performance Index) calculation unit 23 calculates an SPI value (an example of a progress evaluation index) as an evaluation index related to the progress of the process to be processed.
[0056] Here, consider a case where a delay occurs in the progress results at the midpoint of the process for the above-mentioned progress plan shown in Fig. 3. In this case, the actual result SPI calculation unit 23 calculates the actual result period (here, X2) from the reference point of the process (here, the project start date) to the point at which a predetermined progress level (here, Y1) is achieved (the midpoint of the process in the project) based on the data of the actual progress (see the solid line) at the midpoint (time point when the period is X2), as shown in Fig. 4, for example. The actual result period corresponds to the execution time corresponding to the progress level Y1 in the actual progress.
[0057] Furthermore, the actual SPI calculation unit 23 calculates a planned period (here, X1) until the same progress level as the progress level corresponding to the actual period (here, Y1) is achieved based on the progress plan data (see the dashed line in FIG. 3). The planned period corresponds to the execution time corresponding to the progress level Y1 in the progress plan.
[0058] Furthermore, the actual SPI calculation unit 23 calculates the SPI value at the midpoint between the actual period and the planned period.
[0059] The actual SPI calculation unit 23 can calculate the SPI value as the ratio of the planned period to the actual period, for example, based on the following formula (1). SPI value = X1 / X2 (1) X1: Planning period X2: Performance period
[0060] If a project is progressing as planned at the midpoint of the process (when the SPI value is calculated), the actual period will be equal to the planned period, and the SPI value will be 1. If the process is progressing faster than planned at the midpoint, the SPI value will be > 1, and if there is a delay in the process as shown in Figure 4, the SPI value will be < 1.
[0061] The progress prediction unit 24 generates data of the predicted progress result by predicting the progress result after the midpoint based on the SPI value. For example, the progress prediction unit 24 can predict the progress result after the midpoint (the progress of the process and the corresponding execution time) by assuming that the SPI value calculated at the midpoint will remain constant in the future.
[0062] The progress prediction unit 24 can predict the progress results after the midpoint (here, period=X2) at which the SPI value was calculated in the above-mentioned FIG. 4, for example, as shown in FIG. 5 (see the dashed line).
[0063] Furthermore, the progress prediction unit 24 judges whether or not a delay has occurred in the project (process) based on the data of the progress performance after the predicted midpoint. A delay in the project can be judged, for example, by whether or not the completion date of the process in the predicted progress performance exceeds a preset reference completion date. As such a reference completion date, the completion date of the process to be processed in the progress plan may be used, or a date obtained by adding a predetermined allowance period (a period during which delay is allowed) to the completion date of the process to be processed in the progress plan may be used.
[0064] When the progress prediction unit 24 determines that a delay has occurred in the project, it further calculates the productivity of the project. Productivity is the ratio of output to input to the process being processed at the midpoint (productivity = output / input). As input, the man-hours (manpower x working hours) put into the process can be used. As output, the weight, length, or volume of a completed construction (including a partially completed construction) converted into man-hours can be used.
[0065] When the calculated productivity is 1 (or an approximate value thereof), the progress prediction unit 24 determines that it is necessary to add resources required for the project so as to increase the SPI value. In the case of a construction project, the resources include, for example, the number of workers and construction machines. On the other hand, when the calculated productivity is less than 1, the progress prediction unit 24 determines that it is necessary to consider revising the progress plan (including changing the construction method). The progress prediction unit 24 can transmit the judgment result as to whether or not a delay has occurred in the project and the calculated productivity value to the user terminal 3. This enables the user to recognize that it is necessary to add resources or revise the progress plan.
[0066] When the progress prediction unit 24 determines that additional resources are necessary, the resource calculation unit 25 calculates the resources required for the processes after the midpoint (i.e., the resources to correct the delay) and changes the resources set in the progress plan (i.e., sets the additional input of resources).
[0067] The resource calculation unit 25 can calculate the required resources by using a resource calculation model 32 (an example of a second machine learning model) which is a learning model obtained by machine learning. The resource calculation model 32 is obtained by performing machine learning using teacher data indicating the relationship between the resources input to each process in the past performance data and the construction period of the process after the resources are input (including the number of days the execution time is brought forward due to the input of the resources). The resource calculation model 32 is stored in advance in the storage unit 12 as one of the learning models 30 available to the progress prediction server 2.
[0068] The progress prediction unit 24 can calculate a corrected SPI value (an example of a corrected evaluation index) based on the resources changed by the resource calculation unit 25. Furthermore, the progress prediction unit 24 can generate corrected predicted data of the progress result by re-predicting the data of the progress result after the midpoint based on the corrected SPI value. As a result, the corrected data of the progress result progresses so that the construction period (period until the completion date) is shortened from the data of the progress result predicted in the above-mentioned FIG. 5 (see the dashed line) as shown in FIG. 6 (see the thick solid line). As a result, the project manager can complete the process by the preset standard completion date of the project by changing (adding) the resources at the midpoint. Furthermore, the resource calculation unit 25 can be configured to calculate the additional cost required to input the additional resources together with the additional resources required to complete the process by the standard completion date of the project. In this case, the user can compare and consider the cost required for the additional resources and the risk of delaying the completion date of the project, and select whether or not to input the additional resources to correct the project progress.
[0069] The resource calculation unit 25 can transmit information on the necessary resources to the user terminal 3 in order to prompt the user to change the actual resources. At this time, the resource calculation unit 25 can transmit screen data on the corrected predicted data of the progress result as shown in Fig. 6 to the user terminal 3. This allows the user to confirm the usefulness of adding resources from the screen on the predicted data displayed on the display of the user terminal 3. Thereafter, the user can quickly change the resources (add the necessary resources) to obtain a progress result similar to the corrected progress result shown in Fig. 6.
[0070] The storage unit 12 can be configured with hardware such as a storage for storing data and information required for the processing of the progress prediction server 2.
[0071] The communication unit 13 may be configured by hardware including an antenna, a communication circuit, and the like for the progress prediction server 2 to communicate with the user terminal 3, etc. via the network 4. The control unit 11 may also function as a communication control unit that controls communication between the communication unit 13 and the user terminal 3, etc.
[0072] The progress prediction server 2 is composed of a computer equipped with known hardware. The progress prediction server 2 is appropriately equipped with known hardware such as one or more processors, memory, a display, an input device, a network interface, and storage. At least a part of the functions of each of the units 21-25 in the control unit 11 can be realized by the processor executing a predetermined control program (an example of a progress prediction program). Note that in the progress prediction system 1, at least a part of the functions of the progress prediction server 2 described above may be realized by multiple computers working together.
[0073] In the progress prediction system 1, a user can operate the progress prediction server 2 via a user terminal 3. However, the progress prediction server 2 may also be operated directly by a user. In that case, the progress prediction server 2 may be a standalone system, and the user terminal 3 may be omitted.
[0074] FIG. 7 is a flow diagram showing the flow of progress prediction processing by the progress prediction server 2 according to the first embodiment.
[0075] In the progress prediction process, first, the progress prediction server 2 acquires progress plan data and actual progress data at the current time (midpoint) for the process to be processed (ST101, ST102).
[0076] Next, the progress prediction server 2 calculates the current SPI value based on the performance period calculated based on the progress performance data and the plan period calculated based on the progress plan data (ST103).
[0077] Next, the progress prediction server 2 predicts future (after the halfway point) progress results based on the progress plan data and the SPI value (ST104). Furthermore, the progress prediction server 2 determines whether or not a change in resources (addition of resources) is necessary based on the prediction result of future progress results (ST105).
[0078] Therefore, if there is no need to change the resources (No in ST105), the progress prediction process ends. On the other hand, if there is a need to change the resources (Yes in ST105), the progress prediction server 2 calculates the required resources (ST106) and instructs the user (user terminal 3) to add resources. In this way, the user executes the change of resources in the project.
[0079] According to the progress prediction system 1 of the first embodiment, at a given point in time (i.e., midpoint) of a process in an ongoing project, actual progress data (i.e., progress including deadlines) from the given point onwards can be predicted by simple processing.
[0080] Second embodiment Fig. 8 is a functional block diagram of the progress prediction server 2 according to the second embodiment. Fig. 9 is an explanatory diagram showing a milestone matching process in the progress plans of two processes included in a project. Fig. 10 is an explanatory diagram showing the influence of a preceding project on a subsequent project. Fig. 11 is an explanatory diagram related to data on the progress results after a resource change in a process.
[0081] In the progress prediction system 1 according to the second embodiment, the same components as those in the first embodiment are denoted by the same reference numerals and detailed description thereof will be omitted. In addition, in the progress prediction system 1 according to the second embodiment, matters that are not specifically mentioned below are the same as those in the first embodiment.
[0082] In the above-described first embodiment, attention was focused on one of the processes included in the project, but when a project includes multiple processes (here, two processes), the progress of the preceding process A (an example of the first process) may affect the progress of the following process B (an example of the second process). Therefore, the progress prediction system 1 according to the second embodiment is applied to a project including such multiple processes.
[0083] The subsequent process B needs to be started after the process A has been started (i.e., after the process A has progressed to a certain extent). In a construction project, for example, the process A is a process related to foundation work, and the process B is a process related to the work of steel frames to be installed on the foundation established in the foundation work. In addition, for example, the process A may be a process related to steel frame work, and the process B may be a process related to the work of piping supported by the steel frames installed in the steel frame work.
[0084] In the progress prediction system 1 (progress prediction server 2), the progress prediction process for the preceding process A can be executed in the same manner as in the above-mentioned first embodiment. On the other hand, the progress prediction process for process B is affected by the progress record of process A, and therefore needs to be executed while taking this effect into consideration.
[0085] As shown in FIG. 8, the progress prediction server 2 according to the second embodiment includes a control unit 11, a storage unit 12, and a communication unit 13, similarly to the first embodiment.
[0086] In the control unit 11, the progress plan data acquisition unit 21, the progress result data acquisition unit 22, and the result SPI calculation unit 23 sequentially execute the same processes as those in the first embodiment for the process A and the process B. Moreover, the progress prediction unit 24 executes the same process as those in the first embodiment for the process A.
[0087] The progress prediction unit 24 can correct the SPI value of process B based on an impact assessment model 35 (an example of a third machine learning model), which is a model that represents the impact of progress in process A on process B, and the predicted data of the progress actual results for process A by the progress prediction unit 24, and predict the progress actual results of process B based on the corrected SPI value.
[0088] In the progress prediction system 1, before the progress prediction process, a preparation phase is executed to obtain an impact assessment model 35. In the preparation phase, for example, a process of associating one or more milestones (predetermined points in the processes) in the progress plans of process A and process B is executed.
[0089] 9, for example, the milestone association process associates a milestone B1 at the start of process B (period=X1, progress=Y(B1)) with a milestone A1 at the midpoint of process A (period=X1, progress=Y(A1)) corresponding to the milestone B1 (i.e., the execution times overlap). Also, for example, a milestone B2 at the midpoint (period=X2, progress=Y(B2)) corresponding to an inflection point in the progression (curve) of the progress of process B is associated with a milestone A2 at the midpoint (period=X2, progress=Y(A2)) corresponding to the milestone B2.
[0090] Such milestone matching can be performed, for example, according to the milestone matching adopted in the same or similar past projects. In particular, the milestone matching can be performed according to the milestone matching adopted in the past projects that were successfully executed without delays beyond the allowable range. In particular, such past projects can be the projects that had the shortest time from the start of the project to its completion (construction period).
[0091] Next, in the preparation phase, an impact assessment model 35 is generated that represents the impact of changes in progress (delay in deadline, etc.) in milestones A1 and A2 (examples of first milestones) in process A on the corresponding milestones B1 and B2 (examples of second milestones) in process B. The impact assessment model 35 can be understood as a model that identifies the degree of SPI value propagation that calculates how the SPI value of the preceding process A propagates to the SPI value of the following process B. The impact assessment model 35 is stored in the storage unit 12 as one of the learning models 30 that the progress prediction server 2 can use.
[0092] The impact assessment model 35 may be a mathematical model or a machine learning model. For example, the impact assessment model 35 may be composed of a mathematical model. In such a mathematical model, for example, the SPI value of process B (i.e., the subsequent process) related to the milestone to be evaluated may be calculated based on the SPI value obtained in process A (i.e., the preceding process) which is proceeding in parallel with process B. In this case, the SPI value of process B may be a value obtained by multiplying the SPI value of process A by a predetermined coefficient, or the square root of the SPI value of process A. Alternatively, in the mathematical model, when process B is started without waiting for the corresponding milestone of process A (i.e., earlier than the scheduled start date) despite the delay of process A, the SPI value of process B may be calculated based on the planned period of process B and the accelerated period. In this case, the SPI value of process B may be calculated as the ratio of the planned period of process B to the value obtained by adding the accelerated period to the planned period of process B.
[0093] The impact assessment model 35 is obtained, for example, by performing machine learning using training data indicating the relationship between data on each milestone of a preceding process (corresponding to process A) in past performance data (data on the progress level and the corresponding execution time in the progress plan and progress performance) and data on each corresponding milestone of a subsequent process (corresponding to process B). More specifically, the impact assessment model 35 can be obtained as follows.
[0094] For example, consider the case where milestone B1 of process B, which corresponds to milestone A1 (period = X1) of process A in the progress plan, started earlier than planned in the actual progress (started at period = X1 (B)) as shown in Figure 10. The data of the progress plans for processes A and B in Figure 10 is the same as the data shown in Figure 9.
[0095] In Figure 10, at the start of Process B (period = X1(B) (an example of the second milestone period)), Process A has not yet reached Milestone A1 (period = X1(A) (an example of the first milestone period)) (for example, the foundation for the steelwork in Process B has not been completed), so the progress of work in Process B is hindered. As a result, Process B will take longer than planned, resulting in extra resources being used and significant delays in the progress of Process B.
[0096] In this case, the SPI values for each milestone A1, A2, B1, and B2 in Process A and Process B are calculated as follows: SPI value (milestone A1) = X1 / X1(A) SPI value (milestone A2) = X2 / X2(A) SPI value (milestone B1) = X1 / X1(B) SPI value (milestone B2) = X2 / X2(B)
[0097] The impact assessment model 35 can be obtained by performing machine learning using training data showing the relationship between the SPI values for milestones A1 and A2 in process A as described above and the SPI values for the corresponding milestones B1 and B2 in process B.
[0098] In addition, in FIG. 10, the difference ΔX (deviation between plan and actual results) between the execution timing of milestones A1 and A2 in the progress results of process A and the execution timing of milestones B1 and B2 in the corresponding progress results of process B is calculated as follows: ΔX(milestones A1, B1)=X1(B)-X1(A) ΔX(milestones A2, B2)=X2(B)-X2(A)
[0099] The impact assessment model 35 can also be obtained by performing machine learning using training data that shows the relationship between the difference ΔX in the execution timing of milestones A1, A2 in process A and milestones B1, B2 in process B as described above, and the SPI values of the corresponding milestones B1, B2 in process B.
[0100] In addition, when using past performance data for machine learning, the appropriateness of the setting of each milestone may also be taken into consideration (i.e., only past performance data in which milestones are appropriately set may be the subject of learning).
[0101] The progress prediction unit 24 can correct the SPI value of the B process using the progress result data (including predicted progress result data) for the A process and the impact assessment model 35. Furthermore, the progress prediction unit 24 can generate predicted progress result data by predicting the progress result after the midpoint of the B process based on the corrected SPI value. As a result, the corrected progress result data transitions so that the construction period (period until completion date) is shortened after the midpoint of the B process (period=X2(B)) as shown in FIG. 11 (see the thick solid line). As a result, the progress prediction system 1 makes it possible to complete the B process by the reference completion date.
[0102] In the example shown in Figure 11, the progress record of process A has not been modified, but if the progress record of process A is modified (i.e., if additional resources are input), the SPI value of process B, which is modified by impact assessment model 35, also changes in accordance with the modification of the progress record.
[0103] In the progress prediction system 1 according to the second embodiment, when the progress plan data acquisition unit 21 acquires the progress plan data for the B process, the A process may have already started and may be delayed. In such a case, the start time (second milestone) of the process in the progress plan data for the B process will be set to a time earlier than the corresponding first milestone of the A process.
[0104] Therefore, when the progress plan data acquisition unit 21 determines, based on the data of the progress results of the A process, that the A process is delayed and has not reached the first milestone, it can correct the acquired data of the progress plan for the B process so that the start time (second milestone) of the B process is after the first milestone of the corresponding A process. In this case, the start time of the B process is delayed, and the scheduled completion date of the entire project may be delayed from the initial plan, but by executing the B process in accordance with the corrected progress plan (i.e., by keeping the SPI value of the B process constant), it is possible to suppress the extension of the construction period required for the B process itself and prevent excessive cost expenditure.
[0105] FIG. 12 is a flow diagram showing the flow of progress prediction processing by the progress prediction server 2 according to the second embodiment.
[0106] In the progress prediction process, the progress prediction server 2 executes steps ST201-ST203 which are similar to steps ST101-ST103 shown in FIG.
[0107] Next, the progress prediction server 2 judges whether the processing target is a subsequent process (whether it is affected by the preceding process) (ST204). If the processing target is a subsequent process, the progress prediction server 2 corrects the SPI value calculated in step ST203 based on the data of the progress results predicted for the preceding process and the impact assessment model 35 (ST205). On the other hand, if the processing target is not a subsequent process (not affected by the preceding process), the progress prediction server 2 uses the SPI value calculated in step ST203 as it is.
[0108] Thereafter, the progress prediction server 2 executes steps ST207-ST209 which are the same processes as steps ST105-ST107 shown in FIG.
[0109] According to the progress prediction system 1 of the second embodiment, when a project includes multiple processes (here, process A and process B), data after a predetermined point (midpoint) of process B is predicted based on the SPI value corrected on the basis of the impact assessment model that represents the impact between the processes. Therefore, in the subsequent process B, which is affected by the progress performance of the preceding process A, progress performance data after a predetermined point can be predicted by simple processing.
[0110] Although the description of the specific embodiment has been completed above, the present invention is not limited to the above embodiment or modified example, and can be modified in a wide range of ways. The components of the project progress forecast method, progress forecast device, and progress forecast program shown in the above embodiment are not necessarily all essential, and at least those skilled in the art can select them as appropriate without departing from the scope of the present invention. [Explanation of symbols]
[0111] 1: Progress forecast system 2: Progress prediction server (an example of a progress prediction device) 3: User terminal 4: Network 11: Control unit 12: Storage part 13: Communications Department 21: Progress plan data acquisition section 22: Progress performance data acquisition section 23: Actual SPI calculation section 24: Progress forecast section 25: Resource calculation section 28: Project Data 30: Learning model 31: Progress plan acquisition model (an example of the first machine learning model) 32: Resource calculation model (an example of the second machine learning model) 35: Impact assessment model (an example of the third machine learning model)
Claims
1. A project progress forecasting method for forecasting progress of a second step in a project including a preceding first step and a subsequent second step, comprising: the first step includes a first milestone; the second step includes a second milestone that corresponds in time to the first milestone; A project progress forecasting method, comprising: inputting data on the actual progress of the first milestone and data on the actual progress of the second milestone into an impact assessment model that represents the impact of the relationship between the progress of the first milestone and the progress of the second milestone on the progress of the second process, thereby outputting predicted data on the actual progress of the second process.
2. A method for predicting project progress as described in claim 1, wherein the first milestone and the second milestone are determined based on data on the progress of multiple completed projects.
3. A project progress prediction method as described in claim 1, wherein the impact assessment model is determined based on data on the progress of multiple completed projects.
4. 2. The project progress forecasting method according to claim 1, wherein the impact assessment model is a machine learning model that learns the relationship between the progress performance data of the first milestone and the progress performance data of the second milestone and the progress performance data of the second process.
5. A progress prediction device for predicting the progress of a second step in a project including a preceding first step and a subsequent second step, comprising: the first step includes a first milestone; the second step includes a second milestone that corresponds in time to the first milestone; an impact assessment model that represents an impact of a relationship between the progress of the first milestone and the progress of the second milestone on the progress of the second step; A progress prediction device in which the impact assessment model receives as input progress data for the first milestone and progress data for the second milestone, and outputs predicted progress data for the second process.
6. A progress prediction program for predicting the progress of a second step in a project including a preceding first step and a subsequent second step, comprising: the first step includes a first milestone; the second step includes a second milestone that corresponds in time to the first milestone; On the computer, A progress prediction program that outputs predicted data on the progress of the second process from an impact assessment model that represents the impact of the relationship between the progress of the first milestone and the progress of the second milestone on the progress of the second process by inputting data on the progress of the first milestone and data on the progress of the second milestone into the impact assessment model.
7. A method for creating an impact assessment model that represents the impact of a relationship between a first milestone of a first step and a second milestone of a second step on the progress of the second step in a project including a preceding first step and a subsequent second step, comprising: The computer Acquire data on progress results of the first process, including progress results of the first milestone, and data on progress results of the second process, including progress results of the second milestone, for a plurality of completed projects; A method for creating an impact assessment model, which comprises constructing a model that expresses the relationship between the progress performance of the first milestone and the progress performance of the second milestone and the progress performance of the second process, and creating the model as the impact assessment model.